Cooperation among Multiple Medical Institutions on Breast Cancer Survival Status Prediction Based on Federated Learning and Monte Carlo Trees Search

Abstract

In recent years, breast cancer has become a major threat to women's health worldwide, and its incidence has increased significantly with changes in lifestyle and an aging society. According to data released by the World Health Organization in 2024, breast cancer has surpassed lung cancer as the most common malignant tumor in the world. In China, the incidence of breast cancer continues to grow at a rate of 3% to 4% per year, which places higher demands on early diagnosis, treatment plan optimization, and survival prediction of breast cancer. The vigorous development of machine learning technology has brought new opportunities for breast cancer survival prediction. However, in the context of multi-institutional cooperation, data privacy protection has become an important issue to be addressed. This study focuses on this challenge and innovatively introduces the federated learning (FL) paradigm to enable collaborative model training among multiple institutions without disclosing the privacy of local data, effectively breaking the phenomenon of data silos. At the same time, the Monte Carlo tree search (MCTS) algorithm is used to deeply analyze the survival status transition path, providing a more valuable reference basis for medical decision-making. To address the problem of data imbalance, SMOTE technology is used to pre-process the dataset to ensure the effectiveness of model training. The performance of the classification model is comprehensively evaluated using the confusion matrix and the ROC curve. The experimental results show that this method has advantages in terms of prediction accuracy and interpretability, providing a new perspective for breast cancer research.

Ähnliche Arbeiten